Upkeep

Upkeep is a self-hosted uptime and health monitor for personal and portfolio projects running on free-tier hosts (Render, Railway, Fly.io, Vercel, and similar).

Free-tier hosts spin down idle services. The next visitor to a live demo eats a slow cold start, or the project just looks broken. Upkeep pings each project’s health endpoint on a schedule, keeps services warm, and gives you one dashboard instead of N different host dashboards to check.

What it does

• Health checks — periodically checks each project using one of four check types (HTTP, TCP, DNS, SSL certificate) and classifies the result (up, down, degraded, waking, unknown). See Projects & Health Checks for what each type verifies.
• Keep-alive pings — optionally pings a project every 10 minutes (with an optional daily active window) purely to stop a free-tier host idling out — separate from monitoring and never opens an incident.
• Incidents — opens an incident when a project goes down, tracks how long it stays down, and resolves it automatically once checks recover.
• Notifications — alerts a Discord webhook, email, or generic webhook when a project’s status changes, with configurable escalation thresholds and optional daily/weekly digests instead of per-incident noise.
• Dashboard — response-time graphs, uptime heatmaps, and a searchable incident history across every project you own.
• Public status pages — opt any project into an unauthenticated, read-only status page, plus one aggregate portfolio page listing every project you’ve made public. See Public Status Pages.
• Programmatic registration — register a project from a CI/CD deploy step using a per-user API key instead of the dashboard form. See API Keys.

Stack

Upkeep’s backend is Supabase (Postgres, Edge Functions, pg_cron scheduled triggers) and its frontend is a Next.js dashboard, deployable independently of the backend on Vercel, Netlify, or similar.